The Human Homologue of the Yeast Proteins Skb1 and Hsl7p Interacts with Jak Kinases and Contains Protein Methyltransferase Activity*

To expand our understanding of the role of Jak2 in cellular signaling, we used the yeast two-hybrid system to identify Jak2-interacting proteins. One of the clones identified represents a human homologue of the Schizosaccaromyces pombe Shk1 kinase-binding protein 1, Skb1, and the protein encoded by theSaccharomyces cerevisiae HSL7 (histone synthetic lethal 7) gene. Since no functional motifs or biochemical activities for this protein or its homologues had been reported, we sought to determine a biochemical function for this human protein. We demonstrate that this protein is a protein methyltransferase. This protein, designated JBP1 (Jak-binding protein 1), and its homologues contain motifs conserved among protein methyltransferases. JBP1 can be cross-linked to radiolabeled S-adenosylmethionine (AdoMet) and methylates histones (H2A and H4) and myelin basic protein. Mutants containing substitutions within a conserved region likely to be involved in AdoMet binding exhibit little or no activity. We mapped the JBP1 gene to chromosome 14q11.2–21. In addition, JBP1 co-immunoprecipitates with several other proteins, which serve as methyl group acceptors and which may represent physiological targets of this methyltransferase. Messenger RNA for JBP1 is widely expressed in human tissues. We have also identified and sequenced a homologue of JBP1 in Drosophila melanogaster. This report provides a clue to the biochemical function for this conserved protein and suggests that protein methyltransferases may have a role in cellular signaling.

The role of the Jak kinases in cytokine signal transduction was first shown for the interferons (IFNs) (24,25). Subsequently, many reports have demonstrated that Jak activation occurs rapidly after ligand stimulation (1,8,9,26). This activation initiates a cascade of events, which includes receptor phosphorylation and recruitment, subsequent phosphorylation and nuclear translocation of members of the Stat (signal transducers and activators of transcription) family of proteins, which then activate cytokine-inducible genes (27). In addition to their enzymatic role, several reports have demonstrated that the Jaks play a structural role in the receptor complex and that the Jaks may have functions in addition to their kinase activity that are important for signaling. For example, introduction of a kinase-inactive mutant of Jak1 into cells that lack this kinase (and are unresponsive to interferon-␥ (IFN-␥) restores partial IFN-␥-induced gene expression (28,29). Furthermore, the amino terminus of Tyk2 stabilizes the IFNAR1 chain of the IFN-␣ receptor complex (30).
In addition to their interactions with cytokine receptor chains, a large body of evidence has accumulated demonstrating that the Jak kinases interact with other signaling proteins. In particular, Jak2 was reported to interact with SHPTP1, SHPTP2, PP2A, PI3K, Yes, Fyn, Shc, Syp, Grb2, the angiotensin II AT1 receptor, and the serotonin 5-HT2A receptor (31)(32)(33)(34)(35)(36)(37)(38)(39)(40)(41)(42)(43)(44). The ability to interact with such diverse proteins underscores the complex role of Jak2, which is activated by the majority of ligands that utilize the Jak-Stat pathway (45). While the physiological roles for these interactions have not been characterized, they suggest that the Jaks play a role in other pathways and/or facilitate cross-talk between signaling pathways.
In identifying Jak2-interacting proteins with the yeast twohybrid system, we cloned a human homologue of the Schizo-saccaromyces pombe Skb1 protein and the Saccharomyces cerevisiae protein encoded by the HSL7 gene (46 -48). The skb1 gene was initially identified during a two-hybrid screen for proteins interacting with the shk1 kinase which represents a member of the p21 cdc42/Rac1 -activated kinase (PAK) family of protein kinases (47). Recent data suggest that removal of this protein results in cell cycle abnormalities and that the human homologue of this protein can functionally substitute for Skb1 (49). 2 The HSL7 (histone synthetic lethal 7) gene was initially identified as a gene whose mutation is lethal in combination with a mutation in the histone H3 and was described to be a negative regulator of Swe1 function (48). Disruption of HSL7 also results in cell cycle abnormalities of S. cerevisiae (48). Taken together these data suggest that this family of proteins is involved in coordinating cellular events such as the cell cycle or cellular signaling. Since no functional motifs or biochemical activities had been identified for Skb1 or Hsl7p, we focused on identifying a biochemical activity for JBP1. This report shows that JBP1 is a protein methyltransferase.

MATERIALS AND METHODS
Creation of Plasmid Encoding GAL4 DBD -Jak2-Expression vectors for all the Jak kinases (Jak1, Jak2, Jak3,and Tyk2) were gifts from James Ihle and John Krowlewski. The two-hybrid system vector pAS2 which contains the yeast tryptophan (TRP1)-selectable marker and a hemagglutinin (HA) tag, was a gift from Stephan Elledge (50). To create pAS2-Jak2, the murine Jak2 cDNA was modified in the following manner. First, the BsrGI site in the 3Ј-untranslated region of the plasmid BluescriptSK-muJak2 (a gift from James Ihle) was removed by digesting with NheI, blunt ending with the large fragment of DNA polymerase I (Klenow fragment), digesting with EcoRV,and recircularizing with T4 DNA ligase. This created the plasmid muJak2-BsrGI. A linker containing a SfiI site was placed into the remaining BsrGI site located 53 base pairs downstream from the translational start codon. This linker was created by annealing two oligonucleotides (5Ј-GTACGGCCATGGAG-GCC-3Ј and 5Ј-GTACGGCCTCCATGGCC-3Ј), then heating equimolar amounts to 100°C, and cooling slowly to 4°C in 10 mM Tris⅐Cl (pH 7.8), 10 mM MgCl 2 . The annealed linker was ligated into the BsrGI-digested muJak2-BsrGI plasmid. This created the plasmid muJak2-SfiI. The Jak2 cDNA was cloned from muJak2-SfiI into pAS2 as a SfiI-SalI fragment. This created the plasmid pAS2-Jak2, which encoded the GAL4 DNA binding domain fused to amino acids 19 -1129 of murine Jak2.
Yeast Two-hybrid System and Cloning of JBP1-All two-hybrid system materials except the two-hybrid system HeLa cell cDNA library were gifts from Stephan Elledge (50). The HeLa library in vector pGADH, which contains the LEU2 selectable marker, was a gift from Greg Hannon (51). The GAL4 activation domain was fused to the HeLa cell cDNAs in this library. Yeast strains used were Y190 (MATa gal4 gal80 his3 trp1-901 ade2-101 ura3-52 leu2-3,-112 ϩURA3:: GAL3lacZ, LYS2::GAL(UAS)3 HIS3 cyh r ) and Y187 (MAT␣ gal4 gal80 his3 trp1-901 ade2-101 ura3-52 leu2-3,-112 met-URA3::GAL3lacZ). Yeast transformations were performed via the lithium acetate method as described previously (52). The transformed yeast cells were plated onto agar media lacking leucine, tryptophan, and histidine and containing 30 mM 3-aminotriazole to isolate colonies containing both plasmids and activating the HIS3 reporter gene. After 7 days at 30°C, 143 histidine prototrophs were selected and restreaked onto media lacking leucine and tryptophan. These were assayed for ␤-galactosidase expression with the paper filter assay (52). Colonies that showed evidence of ␤-galactosidase expression were streaked onto media lacking leucine and containing 10 g/ml cycloheximide to select for loss of the pAS2-Jak2 plasmid. Colonies from these plates were restreaked onto media lacking leucine and tryptophan to confirm loss of the pAS2-Jak2 plasmid. One to three colonies of each isolate containing the library plasmid without the pAS2-Jak2 plasmid were each mated to yeast strain Y187 containing the plasmid pAS1-CDK2, pAS1-SNF1, pAS1-lamin, or pAS2-Jak2. Diploids from these matings were assayed for ␤-galactosidase expression by the paper filter method. Plasmid DNA from the clones showing evidence of ␤-galactosidase expression only in the presence of Jak2 were rescued, characterized by restriction endonuclease digestion, and sequenced. The two plasmids corresponding to JBP1 were initially called p31-2B and p41-3A and had inserts of approximately 1500 base pairs (46).
To isolate full-length clones corresponding to the cDNA fragments within plasmids p31-2B and p41-3A, the insert from plasmid p41-3A was released with the EcoRI and XhoI restriction endonucleases. This fragment was labeled by the random hexamer method and used to identify hybridizing sequences in a human cDNA library (53,54). Positive phage plaques were purified and plasmid DNA rescued by digesting total DNA with either NotI or MluI restriction endonucleases, recircularizing with T4 DNA ligase and transforming into Escherichia coli. Inserts were sequenced with vector oligonucleotides, internal oligonucleotides, and fragment cloning into pBluescript SKϪ (Stratagene). The largest cDNA isolated was in phagemid p41-6. The insert from phagemid p41-6 was cloned as a SalI fragment into the compatible XhoI site of pBluescript SKϪ to yield plasmid p41-6-Bluescript.
Construction of Expression Vectors-The plasmid pcDNA3-HA was created by amplifying a triple influenza HA tag from vector pSM491 which was a gift from Terry Goss Kinzy (55). This was accomplished by polymerase chain reaction (PCR) with the following oligonucleotides: HA-5Ј (5Ј-GCCGGTACCATGGGCCGCATCTTTTACCCA-3Ј) and HA-AS (5ЈACCACGGAGCCTTTGCTGGAGCGGCCGCACTGAGCAG-CGT-3Ј). Thirty cycles of PCR were used under the following conditions: 20 M of each oligonucleotide, 0.025 g of template plasmid DNA, 0.5 units of Taq DNA polymerase, with buffer containing 10 mM Tris⅐HCl, 50 mM KCl, 1.5 mM MgCl 2 (pH 8.3 at 20°C) in a total volume of 0.05 ml. Thirty cycles at 94°C for 30 s, 50°C for 45 s, and 72°C for 45 s were performed. The product from this reaction was digested sequentially with NotI and KpnI restriction endonucleases and cloned into vector pcDNA3 (Invitrogen), which was digested with NotI and KpnI restriction endonucleases. To create plasmid pEF2-HA, plasmid pcDNA3-HA was digested with KpnI and SfuI restriction endonucleases and the fragment containing the HA tag and multiple cloning site was cloned into vector pcDEF3 digested with the same enzymes (56).
The plasmid to express HA-tagged JBP1 was constructed in three steps. First, PCR was used to amplify the insert with a 5Ј NotI site whose frame was compatible with the NotI site of vector pcDNA3-HA. This was accomplished by PCR with the plasmid p41-6-Bluescript as the template and the following 3Ј and 5Ј oligonucleotide primers, respectively: U2-5-4 (5Ј-TTGTGCCACCACATCCACGT-3Ј) and HA-41-5Ј (5Ј-CGGAATTCGCGGCCGCGCGGTCGGGGGTGCTGGTGG-3Ј). The PCR conditions were the same as above. This PCR product was digested with NotI and NcoI restriction endonucleases and cloned into plasmid p41-6-Bluescript digested with the same enzymes. This created plasmid p41-N, which contained the 41-6 cDNA with a modified 5Ј end containing a NotI site compatible with vector pcDNA3-HA. This modified cDNA was cloned into pcDNA3-HA as a NotI /ApaI fragment, creating plasmid pcDNA3-HA-41-6 also called plasmid pcDNA3-HAJBP1. To create plasmid pEF2-HAJBP1, the plasmid pcDNA3-HAJBP1 was digested with NotI and AvrII restriction endonucleases and the fragment containing the insert was cloned into plasmid pEF2-HA digested with the same enzymes.
The pEF2-ATG-FLAG vector was created by inserting a DNA fragment encoding the Flag epitope into vector pcDEF3 (56). The DNA fragment encoding the Flag sequence was created with PCR using the plasmid pFL␥R2 (57) as the template and the following primers, 5Ј-GGGGTACCTATGGACTACAAGGACGACGAT-3Ј and 5Ј-GTCTGGCG-GATCCGCCTTGTC-3Ј. The amplified fragment was digested with the BamHI and KpnI restriction endonucleases and cloned into vector pc-DEF3 digested with the same enzymes (56). The plasmid pEF2-ATG-FLAG-N vector was created by digesting the plasmid pEF2-ATG-FLAG with the BamHI restriction endonuclease and inserting the self-annealed oligonucleotide 5Ј-GATCGCGGCCGC-3Ј into the digested BamHI site. The plasmid pEF2-FLAG-JBP1 was created by digesting the plasmid pEF2-ATG-FLAG-N with the NotI and XbaI restriction endonucleases and inserting a NotI/XbaI DNA fragment containing the JBP1 cDNA from the plasmid pBSSK-HA-JBP1. The plasmid pBSSK-HA-JBP1 was created by digesting the plasmid pcDNA3-HA-JBP1 with the AseI restriction endonuclease, blunt-ending with the large fragment of DNA polymerase I (Klenow), and inserting the fragment containing the HA-JBP1 insert into the plasmid pBluescriptSK (Stratagene) digested with the restriction endonuclease EcoRV and dephosphorylated with alkaline phosphatase. The vector pEF2-ATG-Myc was created by inserting a DNA fragment encoding the Myc epitope into vector pcDEF3 (56). The DNA fragment encoding the Myc sequence was created with PCR using a plasmid containing the Myc epitope as the template (a gift from Jerry Langer) and the following primers, 5Ј-GGGGTACCATG-GAAGAGCAGAAGCTGATC-3Ј and 5Ј-CCGGATCCAGGTCCTCCTCA-2 S. Marcus, personal communication.
GAGATC-3Ј. The conditions for PCR were the same as above. The amplified fragment was digested with the BamHI and KpnI restriction endonucleases and cloned into vector pcDEF3 digested with the same enzymes (56). The plasmid pEF2-ATG-Myc-N vector was created by digesting the plasmid pEF2-ATG-Myc with the BamHI restriction endonuclease and inserting the self-annealed oligonucleotide 5Ј-GATCGCGGCCGC-3Ј into the digested BamHI site. JBP1 was cloned into pEF2-ATG-Myc-N as a NotI/XbaI fragment from pEF2-Flag-JBP1.
Construction of pEF2-HAJBP1D4 -The plasmid pEF2HAJBP1D4 was created to express a carboxyl-terminal fragment of the JBP1 protein for purification and deletion analysis. A fragment of the JBP1 cDNA was amplified by PCR with the plasmid p41-6-Bluescript as the template and the T7 oligonucleotide and the oligonucleotide 41-del-4 5Ј-GCGGCCGCGCCCAGCACTTCCTAAAAGATG as primers. The PCR product was cloned into vector pCR2.1 (Stratagene) and then subcloned as a NotI/SpeI fragment into vector pEF2HA digested with NotI and XbaI restriction endonucleases.
Creation of JBP1 Point Mutants-The PCR was used to create two point mutants of JBP1 called JBP1 R368A and JBP1 G367A . To create JBP1 R368A , a fragment of JBP1 was amplified with the oligonucleotides 41Del-1 (5Ј-GCGGCCGCCCCTTGGTGGCACCAGAGG-3Ј) and MUT1 (5Ј-GTTCACCAGGGGTCCCGCTCCTGCTCCC-3Ј; changed bases shown in boldface) and the plasmid pEF2-Flag-JBP1 as the template. Thirty cycles at 94°C for 40 s, 42°C for 40 s, and 72°C for 40 s were performed with the same buffer and concentrations as mentioned above. To create JBP1 G367A , PCR was performed as described for JBP1 R368A except that the oligonucleotide MUT2 (5Ј-GTTCAC-CAGGGGTCCCCGTGCTGCTCCC-3Ј) was used instead of MUT1. After amplification, these fragments were extracted with phenol:chloroform:isoamyl alcohol, precipitated with ethanol, and digested with the restriction endonucleases SanDI and NdeI. The fragments containing the mutations were purified from agarose and ligated into the plasmid pEF2-Flag-JBP1 digested with same restriction endonucleases. To express the JBP1 R368A and JBP1 G367A proteins fused to the Myc epitope, the inserts from the Flag constructs were cloned into the pEF2-ATG-Myc-N vector as NotI/XbaI fragments.
Cell Culture and Transient DNA Transfections-COS-1 cells (58), derived from a simian kidney line, were transfected with plasmids by the DEAE-dextran/Me 2 SO shock protocol (59 -61). Tissue culture dishes (10-cm dish, Falcon) containing 10 ml of 10% calf serum-supplemented Dulbecco's modified Eagle's medium (DMEM) were seeded with 1.75-2.0 ϫ 10 6 cells trypsinized at confluence. Cells were incubated overnight at 37°C before transfecting. For each dish, 636 l of DMEM, 160 l of DEAE-dextran (10 mg/ml, filter-sterilized), and 4 l of chloroquine diphosphate (60 mg/ml, filter-sterilized) were combined. Next, for each dish transfected, 5 g of plasmid DNA was diluted in 200 l of TBS (25 mM HCl, 136 mM NaCl, and 2.7 mM KCl, pH 7.4) and then added dropwise to the 800-l DMEM/DEAE-dextran/chloroquine diphosphate mixture. Medium was aspirated from a dish of cells, then this DNA/DMEM/DEAE-dextran/chloroquine diphosphate mixture was then added to the dish, swirled gently for a few seconds, topped with 3 ml of 2.5% calf serum supplemented DMEM, and incubated for 3 h at 37°C. After the incubation, the medium was removed by aspiration and the cells were shocked by the addition of 3 ml of PBS containing 10% Me 2 SO. After 2 min the Me 2 SO was aspirated, the cells rinsed with 3 ml of PBS, and then returned to the 37°C incubator with 10 ml of complete DMEM. Cells were harvested for assays after 48 -72 h.
Isolation of HeLa Cell Lines Expressing HA-JBP1-HeLa cells were cultured in DMEM supplemented with 10% fetal bovine serum (FBS). Cells from two T-150 flasks were grown to 90% confluence, and adherent cells were trypsinized and transferred to a 15-ml tube. Cells were washed twice with ice-cold PBS and resuspended in 0.45 ml of PBS. Ten g of plasmid DNA, pEF2HA, or pEF2HA-JBP1, was added to the cells and mixed by pipetting. The mixture of cells and DNA was transferred to an ice-cold cuvette with a gap of 0.4 cm. Cuvettes were pulsed with a Bio-Rad Gene Pulser II set at 220 V and 950 microfarads. After elec-troporation, 1 ml of DMEM containing 10% FBS was added and the cells were mixed well by pipetting. The contents of the cuvette were transferred to a 150 mm dish and DMEM containing 10% FBS was added. After 72 h, cells were grown under selection with 450 g/ml G418 (Geneticin). Independent colonies were selected and expanded in separate dishes. Clones were analyzed by Western blotting with anti-HA antibody for expression of HA-JBP1.
In Vitro Binding Assay-To express the recombinant protein fused to glutathione S-transferase (GST), the insert from plasmid p31-2B was cloned into vector pGEX1 (Amersham Pharmacia Biotech) as a BamHI/ XhoI fragment creating plasmid pGST-JBP1. This fused amino acids 268 -637 of JBP1 to GST and is referred to as GST-JBP1-N268. This plasmid was transformed into E. coli strain DH5␣ F'IQ. Induction and purification of the GST-JBP1 fusion protein were performed as follows. Cultures of 250 ml were grown to an OD 600 of 0.8 at room temperature. Cultures were induced with 1 mM IPTG overnight at room temperature. Cells were pelleted by centrifugation at 4700 ϫ g (rotor H6000A, 4,000 rpm) in a Sorvall RC3C centrifuge at 4°C. Cells were washed one time with Buffer A (150 mM NaCl, 50 mM Tris⅐HCl, pH 7.4, and 0.8 mM PMSF), then repelleted and resuspended in 12.5 ml of Buffer A and sonicated. Triton X-100 (20%) was added to a final concentration of 1%, and lysates were incubated at 4°C with agitation for 20 min. Lysates were centrifuged at 4°C for 20 min at 8,720 ϫ g (rotor SS-34, 18,000 rpm) in a Sorvall RC5B centrifuge. Supernatant was divided into 0.8-ml aliquots, frozen in a dry ice ethanol bath, and stored in liquid nitrogen or at Ϫ70°C. To perform the in vitro binding assay, 0.8 ml of frozen lysate was thawed and centrifuged at maximum speed in an Eppendorf 5415 centrifuge for 2 min. The GST control or GST fusion protein (GST-JBP1-N268) was purified by incubating a fraction of this lysate (equivalent to 0.4 -2.0 g of fusion protein) with 50 l of 50% glutathione-agarose (Sigma no. G4510) in Buffer A for 30 min at 4°C with rocking. Afterward, the beads were washed twice in 1 ml of Buffer B (Buffer A containing 500 mM NaCl) for 5 min. This was followed by three additional washes in 1 ml of Buffer C (25 mM Tris⅐HCl, pH 7.4, 50 mM KCl, 10 mM MgCl 2 , 2% glycerol, 0.8 mM PMSF, 0.1% Triton X-100, and 100 g/ml BSA) for 3 min. Buffer C was added to the beads to yield a final volume of 100 l.
The binding reaction was carried out by incubating 25 l of washed beads in Buffer C with lysate from COS cells overexpressing different Jak kinases. After incubation for 1 h at 4°C, the beads were pelleted for 2 min at 500 ϫ g, then washed twice for 2 min with rocking in 1.0 ml of Buffer C. Bound proteins were eluted by adding 25 l of 2ϫ sample buffer (100 mM Tris⅐HCl, pH 6.8, 4% SDS, 0.2% bromphenol blue, 20% glycerol, 200 mM DTT) to the washed beads and boiling for 2 min. Eluted proteins were resolved by SDS-PAGE and transferred to polyvinylidene difluoride (PVDF) membranes (Bio-Rad) with the Trans-Blot SD semidry transfer cell (Bio-Rad, catalog no. 170-3910) according to manufacturer's instructions. Membranes were rinsed in PBS (138 mM NaCl, 2.7 mM KCl, 1.5 mM KH 2 PO 4 , 8.2 mM Na 2 HPO 4 ) containing 0.05% Tween 20 and blocked in a solution containing 3% nonfat dry milk (Shoprite brand), 1% bovine serum albumin, 1 M dextrose, 10% glycerol, 0.05% Tween 20, and 0.05% thimerosal (Sigma no. T8784) for 1-12 h at room temperature. Membranes were probed with the indicated primary antibodies for 1-12 h and washed twice in PBS (138 mM NaCl, 2.7 mM KCl, 1.5 mM KH 2 PO 4 , 8.2 mM Na 2 HPO 4 ) containing 0.05% Tween 20 for 15 min per wash; then incubated for 1 h with the indicated secondary antibody conjugated to horseradish peroxidase followed by washing as above. The bands were visualized with Western blot Chemiluminescence Reagent Plus from NEN Life Science Products (catalog no. NEL105).
Immunoprecipitations from Mammalian Cells-Dishes of adherent mammalian cells were washed twice with ice-cold PBS and scraped with a silicone policeman in 0.5 ml of Lysis Buffer. Cells were incubated on ice for 20 min to 12 h and spun down at 16,000 ϫ g for 10 min at 4°C. Supernatants were transferred to a new 1.5-ml microcentrifuge tube. To immunoprecipitate epitope-tagged JBP1, 0.2 ml of the above lysate was incubated with 0.5-2.0 g of the appropriate antibody for 1 to 12 h with rocking. Protein A/G plus beads (Santa Cruz sc-2003) were added and incubated for 1 h with rocking. The beads were pelleted and then washed three times in 1 ml of ice-cold Lysis Buffer for 1 to 10 min per wash. Bound proteins were eluted with sample buffer and resolved with SDS-PAGE. Proteins were transferred and blotted for Western analysis.
Affinity Purification of Flag-JBP1 and (His) 6 -JBP1D4 -Flag-JBP1 was transiently expressed in COS cells, and cell lysates were prepared as described above for immunoprecipitation. Cell lysate was passed over an anti-Flag M2 affinity column created by packing 1 ml of anti-Flag M2 affinity gel (Eastman Kodak Co. catalog no. IB13020) into a Econo-Pac disposable chromatography column (Bio-Rad catalog no. 732-1010). The column was first washed in 15 ml of ice-cold glycine buffer containing 0.1 M glycine-HCl, pH 3.5, followed by washing with 15 ml of TBS (50 mM Tris⅐HCl, 150 mM NaCl, pH 7.5). Cell lysate was passed over the column by gravity flow three to four times, and the column was washed with 50 ml of ice-cold TBS. Flag-JBP1 was eluted in 1-ml fractions in TBS containing 250 g/ml Flag peptide (Kodak catalog no. IB13070), and fractions were analyzed by SDS-PAGE and Coomassie Blue staining.
To purify recombinant protein, the insert from plasmid pEF2HAJBP1D4 was cloned into the BamHI site of the pRBSII sixhistidine vector of the appropriate frame (94, 95). This fused a sixhistidine epitope tag to the NH 2 terminus of proteins encoded by the JBP1D4 DNA insert. E. coli carrying the appropriate plasmid were induced with 1 mM IPTG for 3-12 h at 37°C. Cells were pelleted at 4651 ϫ g (4000 rpm) in a RC3C Sorvall centrifuge with a H600A rotor. Cells were resuspended in denaturing Lysis Buffer (final pH of 8.0, 50 mM Na 2 PO 4 , 10 mM Tris⅐Cl, 6 M guanidine HCl, 100 mM NaCl) at 0.5 ml/2 ml of liquid culture. The lysate was centrifuged at 14,000 rpm (23, 420 ϫ g) in a Sorvall RC5B centrifuge with an SS-34 rotor. Supernatant was loaded onto a column containing 2 ml of Talon (15), Flag-JBP1, GST, or GST-JBP1-N268 were each incubated with 5.5 Ci of [ 3 H]S-adenosylmethionine (NEN catalog no. NET155H) in cross-linking buffer (50 mM Tris⅐Cl, pH 7.5, 0.1 M NaCl, 2 mM EDTA, 1 mM DTT) in a total volume of 0.065 ml. Samples were added to 96 well plates and incubated on ice at a distance of 3.5-5.0 cm from the UV source (Stratalinker 2400). Samples were exposed to two 0.96 joules of UV irradiation and the reaction stopped by the addition of 30 l of 3ϫ SDS-PAGE sample buffer. Samples were stored at Ϫ20°C overnight, and then proteins were separated by SDS-PAGE and stained with Coomassie Blue. This was followed by incubation with Entensify reagent (NEN catalog no. NEF992) and radiography with Kodak Biomax MR film at Ϫ70°C for 7-14 days.
Methylation of Proteins-HA-JBP1 was immunoprecipitated with anti-HA antibody from the HeLa-HA-SPMT cell line as described above and washed three times for 10 min in 0.5 ml of ice-cold Lysis Buffer. As a control for the immunoprecipitation, HeLa cells transfected with the parent vector pEF2HA containing no insert were used. This was followed by rinsing the immunoprecipitates two times with Methylation Buffer (50 mM Tris⅐HCl, pH 7.5, 1 mM EDTA, and 1 mM EGTA) at room temperature. Protein A/G beads with bound HA-JBP1 were resuspended in 0.04 ml of Methylation Buffer, and either no additional methyl acceptor was added or 10 -100 g of pooled histones (Sigma catalog no. H-9250), individual histones separately (H1, H2A, H2B, H3, or H4 from Roche Molecular Biochemicals; catalog nos. 223549, 1034740, 223514, 1034758, 223492), cytochrome c (Sigma catalog no. C-7752), or myelin basic protein (Sigma catalog no. M1891) was added. To these mixtures, which were placed on ice after washing with Methylation Buffer, 0.005 ml (2.75 Ci) of [ 3 H]AdoMet (NEN catalog nos. Net115 or NET115H) was added. Reactions were incubated at 30°C for 30 min, and reactions were stopped by the addition of SDS-sample loading buffer. Samples were boiled for 2 min and loaded onto 15% SDS-polyacrylamide gels. After running (70 mV through the stacking gel and 140 mV through the resolving gel) the gels were stained in Coomassie Blue, destained, treated with Entensify, and analyzed by radiography.
Northern Analysis-Multiple tissue Northern blots (CLONTECH) were probed with labeled DNA sequences from JBP1. To probe the Northern blots, a 0.7-kb fragment of JBP1 was labeled by the random hexamer method (53). The 0.7-kb JBP1 fragment was created by digesting the JBP1 cDNA with the NdeI and HindIII restriction endonucleases and purifying the fragment from a 1% agarose gel.
Development of Antisera-Antisera were developed in rabbits at Lampire Biological Laboratories (Ottsville, PA). To develop antibodies against the JBP1 protein, the (His) 6 -JBP1D4 protein was purified as described in the section on the purification of JBP1D4 and injected into rabbit 6511. Rabbit 6511 was immunized with 1 mg of the (His) 6 -JBP1D4 protein in 5 ml of PBS (2.5 ml injected intradermally along the back and 2.5 ml subcutaneously into the axilla) on 12/18/97. This was followed on 1/5/98 with 0.52 mg of (His) 6 -JBP1D4 in 4.2 ml (or 1 mg in 7 ml on 1/14/98) of incomplete Freund's adjuvant (0.5 volume injected intradermally along the back and 0.5 volume subcutaneously into the axilla). These injections were followed with intramuscular injections into the hind leg of the (His) 6 -JBP1D4 protein in incomplete Freund's adjuvant as follows: 1 mg in 3 ml on 2/9/98, 1 mg in 7 ml on 3/16/98, 1 mg in 4.5 ml on 4/13/98, 0.83 mg in 9 ml on 5/7/98.
Fluorescence in Situ Hybridization (FISH)-In situ hybridization of JBP1 to human metaphase chromosomes was performed using a modification of the previously published technique (62). A plasmid containing JBP1 was biotin-labeled by nick translation and hybridized to metaphase spreads of human lymphocytes on glass slides. The slides were first pretreated with RNase A (Oncor), washed in 2ϫ SSC, dehydrated, and then denatured at 70°C in 70% formamide, 2ϫ SSC immediately prior to hybridization. The hybridization mixture (15 l/ slide) contained the biotinylated probe (300 ng/slide) and excess repetitive human DNA (Blockit, Oncor) in Hybrisol VI (Oncor). After overnight hybridization at 37°C, slides were washed at 45°C in 50% formamide/1ϫ SSC (2 ϫ 10 min), 1ϫ SSC (2 ϫ 5 min), 0.1ϫ SSC (1 ϫ 5 min), and 0.1ϫ SSC (1 ϫ 5 min) at room temperature. Probe detection was performed by incubation of the slides with fluorescein isothiocyanate-avidin (Oncor) for 20 min at 37°C. Signal amplification was performed by subsequent incubation with an anti-avidin antibody (Oncor), followed by incubation with fluorescein isothiocyanate-avidin. The slides were mounted in the antifade medium (Oncor) containing diamidinophenylindole and analyzed using a BX60 Olympus fluorescence microscope. Color prints of the metaphase spreads showing hybridization with JBP1 were obtained with Cytovision (Applied Imaging). The slides were subsequently destained and Giemsa-banded (63). Determination of Protein Concentration-Proteins were quantitated with the Bio-Rad Protein Assay (catalog no. 500-0006) or by comparing SDS-PAGE samples stained with Coomassie Blue with BSA as the reference protein.

Identification and Cloning of JBP1 with the Yeast Two-hy-
brid System-To screen for Jak2 interacting proteins with the two-hybrid system, we cloned the murine Jak2 cDNA into vector pAS2 (50) as described under "Materials and Methods." This construct fused the GAL4 DNA binding domain (GAL4 DBD ) to amino acids 19 -1129 of the murine Jak2 cDNA.
The yeast strain Y190 was cotransformed with the plasmid pAS2-Jak2 and a HeLa cell library created for use in the two-hybrid system (51). Yeast strain Y190 contains two reporter genes whose transcription indicates an interaction between the two GAL4 fusion proteins. We first selected transformants for their ability to activate the HIS3 reporter gene by plating onto media lacking histidine. Histidine prototrophs were selected and subsequently tested for activation of the lacZ reporter gene. After screening 1.6 ϫ 10 6 transformants, 143 histidine prototrophs were selected to assay for ␤-galactosidase activity. Twenty-eight of those assayed showed evidence of ␤-galactosidase activity. To determine whether the interactions were dependent on expression of the GAL4-Jak2 fusion protein, we performed a mating assay as described previously (50). Transformants activating both reporter genes were cured of the pAS2-Jak2 plasmid by selecting on media containing cycloheximide. The plasmid pAS2 contains a marker that confers cycloheximide sensitivity to the yeast strain Y190, which is cycloheximide-resistant (50). We confirmed loss of the pAS2-Jak2 plasmid by testing the cycloheximide-resistant colonies for tryptophan auxotrophy. Colonies, which now contained only the library plasmid, were mated to yeast strain Y187 expressing "decoy" GAL4 fusion proteins including lamin, CDK2, or SNF1. We also re-tested for the original Jak2 interaction by including yeast expressing the GAL4-Jak2 fusion protein in this mating assay. Of the 28 clones that exhibited ␤-galactosidase activity, 10 activated the lacZ reporter gene in the presence of the GAL4-Jak2 fusion protein, but not in the pres-ence of the "decoys." Results from one of these mating assays can be seen in Fig. 1. Library plasmids from these 10 clones were rescued, characterized by restriction endonuclease digestion, and sequenced.
The 10 clones which activated the lacZ reporter gene only in the presence of the pAS2-Jak2 construct represented four different cDNAs (46). Three of these cDNAs will be described elsewhere. Two of the 10 clones, called 31-2B and 41-3A, represented independent isolates of the same cDNA. The inserts from clones 31-2B and 41-3A were each roughly 1.5 kb. To isolate a longer clone, we used the 41-3A insert as a probe to screen a human M426 cDNA library (54). This led to the isolation of a 2.4-kb cDNA clone, which codes for a protein of 637 amino acids called JBP1 (for Jak-binding protein 1), with a predicted molecular mass of 72.4 kDa. The putative full-length sequence (GenBank accession no. AF167572) was compiled from our sequence with an additional 97 nucleotides from a sequence tag recently entered into the GenBank (accession no. AA417623). This added an additional two amino acids plus some 5Ј-untranslated region nucleotides to our cDNA. The open reading frame of the JBP1 cDNA encoded the amino acid sequence shown in Fig. 2A.
In Vitro Interaction between GST-JBP1 and the Jak Kinases-To determine whether JBP1 could interact with Jak2 outside of the two-hybrid system, we created a GST fusion protein by cloning the insert from the plasmid p31-2B into the pGEX1 expression vector through a compatible BamHI site (see "Materials and Methods"). This fused amino acids 268 -637 of JBP1 to GST and is referred to as GST-JBP1-N268. We then purified and immobilized recombinant GST-JBP1-N268 onto glutathione-agarose beads. To provide a source of the Jak proteins, COS-1 cells were transiently transfected with different Jak kinase expression plasmids. Lysates from these cells were prepared and incubated with immobilized GST or GST-JBP1-N268. Proteins that bound to the beads were eluted by boiling in sample buffer, separated by SDS-PAGE, transferred to PVDF membranes, and immunoblotted with the respective anti-Jak antibodies. GST-JBP1-N268, but not GST alone, was able to bind murine Jak1, murine Jak2, murine Jak3, and human Tyk2 from COS-1 cells (Fig. 3). The interaction (with Jak2) did not appear to require an active kinase domain, as GST-JBP1-N268 was also able to interact with a kinase inactive mutant of Jak2.
Identification of Sequence Homology between JBP1 and Protein Methyltransferases-Initial searches of the GenBank data bases with our 41-3A sequence revealed homology to sequence tags from a variety of different libraries. Subsequent entries led to the identification of JBP1 as a homologue of the protein encoded by the S. pombe gene skb1 and the S. cerevisiae gene HSL7 (47,48). In these reports, homologous open reading frames from an uncharacterized human cDNA sequence were identified (47,48). Since our original report (46), additional sequence data for this human cDNA have been entered into the GenBank data base (accession no. AF015913). This sequence is identical to our JBP1 sequence with the exception of a substitution of valine for glycine at position 553 and phenylalanine for serine at position 247 (Fig. 2). In addition, when we searched the data bases with our sequence, it matched highly to the complement of the 3Ј-untranslated region of a murine cDNA encoding bach protein 2 as well as other murine sequence tags. When these murine sequences were compiled and translated (GenBank accession no. AF167573), they encoded a protein with extremely high homology to our human cDNA as shown in Fig. 2A. We have also obtained a D. melanogaster homologue of JBP1. By searching available data bases, we identified Drosophila expressed sequence tags with homology to JBP1. After obtaining these clones (LD07634 and LD08768) from Genome Systems Inc., they were sequenced with vector and internal primers (GenBank accession no. AF167574). Initial sequencing revealed that clone LD08768 had a deletion of amino acids 126 -130. Whether this represents a physiological splice variant of this protein remains to be seen. Fig. 2A shows the homology between members of the JBP1 family which currently includes sequences or cDNA clones from S. pombe, S. cerevisiae, Caenorhabditis elegans, D. melanogaster, Mus musculus, and Homo sapiens. Since we used the murine Jak2 cDNA as the bait to screen a human library, we anticipated that any proteins identified would be highly conserved between mice and humans as is the case with these proteins.
Other than the open reading frames previously reported (47,48) and the Drosophila and murine homologues of JBP1 reported here, JBP1 shares little similarity to other known proteins and contains no easily recognizable domains. However, continued analysis revealed that JBP1 shared some homology to protein arginine methyltransferases (64 -67). Regions that had been described to be conserved in methyltransferases appeared to be present in JBP1 and its homologues (68,69). These three regions are shown in Fig. 2B. Because of this homology and the fact that methyltransferases have been shown to vary widely in their primary sequence, we determined whether JBP1 exhibited protein methyltransferase activity (70).
Cross-linking of JBP1 to [ 3 H]AdoMet-To determine whether JBP1 represented a new methyltransferase, we first measured whether this protein could bind to the universal methyl donor AdoMet. One method that has been frequently employed to determine AdoMet binding is UV cross-linking (71,72). To express JBP1 in mammalian cells, we cloned this sequence into mammalian expression vectors containing the Flag epitope. This construct fused amino acids 5-637 of JBP1 with the Flag epitope. We purified FLAG-JBP1 from transiently transfected COS cells by affinity chromatography, incubated this protein with [ 3 H]AdoMet, and exposed the reaction mix-

FIG. 1. Specificity of the interaction between Jak2 and JBP1.
Specificity of the interaction between Jak2 and JBP1 in the two-hybrid system. Three independent colonies of yeast strain Y190 containing a library plasmid encoding JBP1 were mated with yeast strain Y187 containing the pAS2-Jak2 plasmid as well as three "decoy" plasmids. The decoy control plasmids contained the GAL4 DNA binding domain (GAL4 DBD ) fused to lamin, CDK2, or SNF1. Diploids were grown on selective media and assayed for expression of ␤-galactosidase with the paper filter assay. Details of the procedure are described under "Materials and Methods."  (47,48). The murine homologue of JBP1 called JBP1MM was obtained by translating the murine bach 2 cDNA sequence (accession no. D86604) and available murine expressed sequence tag sequences. The D. melanogaster homologue of JBP1 was obtained from a clone identified and sequenced as detailed under "Results." The consensus sequence is shown on the bottom of each panel. Identical amino acids corresponding to the consensus sequence are shown in black outline with white lettering. Similar amino acids are shown in gray outline with white lettering. Dots represent spaces inserted to maintain alignment. The sequences for Hsl7p (GenBank accession no. P38274) and the JBP1 homologue in C. elegans (GenBank accession no. P46580) are not shown in their entirety. These sequences can be accessed from the GenBank. Alignment was generated with the Genetics Computer Group (GCG) (93) pileup software and shading generated with Boxshade software available via the World Wide Web. B, comparison of JBP1 and several protein methyltransferases (or putative methyltransferases). Asterisks and Roman numerals indicate methyltransferase regions I, II, and III as described (68,69). Alignment was generated as above and the accession numbers are as follows: PRMT1 (rat) Q63009, PRMT1 (human) Q99873, RMT1 (yeast) P38074, and PRMT2 (human) P55345. Human and rat PRMT1 and yeast RMT1 have been shown to have methyltransferase activity. The gene for PRMT2 has been localized, but no biochemical activity for its encoded protein has been reported. ture to UV light as described under "Materials and Methods." As a positive control, we used the CheR methyltransferase, a well studied bacterial methyltransferase whose crystal structure was determined (70). We included BSA, GST, and GST-JBP1-N268 as additional controls. Fig. 4A shows that of the four proteins tested only CheR and Flag-JBP1 were able to be cross-linked to [ 3 H]AdoMet.
JBP1 Immunoprecipitates Contain Protein Methyltransferase Activity-Given that JBP1 had homology to a protein methyltransferase and was able to be cross-linked to [ 3 H]AdoMet, we determined whether JBP1 could transfer labeled methyl groups from [ 3 H]AdoMet to proteins. HeLa cells were transfected with the plasmid pEF2HA-JBP1, and stable cell lines expressing HA-JBP1 were isolated. We immunoprecipitated HA-JBP1 from HeLa cells and incubated this protein with histones and [ 3 H]AdoMet. Histones were selected because they have previously been shown to function as a methyl acceptor for some protein methyltransferases (73)(74)(75) and because of the genetic link between HSL7 and histones in yeast (48,75). As a control we used immunoprecipitates from HeLa cells transfected with the pEF2-HA vector. Incubations of histones with [ 3 H]AdoMet or [ 3 H]AdoMet plus immunoprecipitates from pEF2-HA vector-transfected HeLa cells resulted in no transfer of the radioactive methyl groups (Fig. 4B). However, incubation of histones with [ 3 H]AdoMet plus immunoprecipitates from transfected cells expressing HA-JBP1 resulted in transfer of methyl groups from [ 3 H]AdoMet to histones. To determine whether HA-JBP1 could transfer methyl groups to other methyl acceptors, we also included two additional known methyl acceptors, cytochrome c and myelin basic protein (74,76). Myelin basic protein, but not cytochrome c, could serve as a methyl acceptor for HA-JBP1.
In addition to reactions containing the exogenously added methyl acceptors, we observed labeled proteins in reactions containing only HA-JBP1 immunoprecipitates. These additional substrates which co-immunoprecipitate with HA-JBP1 vary in molecular weight and may represent additional physiological substrates of JBP1.
Analysis of JBP1 R368A and JBP1 G367A Point Mutants-While we identified homology between JBP1 and putative protein methyltransferases and demonstrated cross linking of JBP1 to AdoMet, the possibility of a contaminating enzyme in our immunoprecipitates could not be excluded. To address this issue, we created two point mutants of JBP1 and analyzed immunoprecipitates of these mutants for protein methyltransferase activity. Because of the highly conserved nature of the GXGRGP motif in JBP1 and its homologues and the similar GXGXG motif in other protein methyltransferases (69), we selected this region for mutational analysis. Both the conserved arginine in JBP1 R368A and the central glycine in JBP1 G367A were mutated to alanine residues. A similar region has been shown to be involved in AdoMet binding of the HhaI DNA methyltransferase (77). We compared the activity of the two point mutants with that of the wild type enzyme after expression in COS cells. While all three proteins were expressed in COS cells, the Myc-JBP1 protein had the ability to methylate histones whereas the mutants exhibited little or no activity as shown in Fig. 4C. Similar quantities of JBP1 and the two mutants were present in the immunoprecipitates as determined by Coomassie Blue staining.
JBP1 Specifically Methylates Histones H2A and H4 -To determine which of the five histones could serve as a substrate for JBP1, methylation reactions were carried out with preparations of individual histones. HA-JBP1 was immunoprecipitated from HA-JBP1-producing HeLa cells as described above. Immunoprecipitates were incubated alone or with 10 g of pooled histones, histone H1, H2A, H2B, H3, H4, myelin basic protein, or cytochrome c (Fig. 5). After separation of the proteins by SDS-PAGE and radiography, it was seen that only histone H2A and H4 were methylated in this experiment. Bands in other lanes were due to the fact that the individual histone preparations were not homogeneous and contained histone H2A or H4 as contaminants. The protein bands corresponding to histones H1, H2B, and H3 were not radiolabeled. The radiographic signals present in the lanes containing histones H1, H2B, and H3 appear to be the same size as histones H2A and H4 (Fig.  5B). These labeled proteins likely represent histones H2A and H4, which were present within these preparations. In addition, labeled proteins of larger molecular weight can be seen in the lanes for histones H2A and H3. These represent other proteins within these preparations which can serve as substrates for JBP1. As shown previously in Fig. 4, myelin basic protein, but not cytochrome c, could also serve as a substrate for JBP1.
HA-JBP1 Interacts with Itself-The homologue of JBP1 was shown to form a homodimer in the two-hybrid system (47). In addition, we observed a doublet in our HA-JBP1 immunoprecipitates after staining with Coomassie Blue even though blotting with anti-HA antibodies revealed only a single band (Figs. 4B and 6). To determine whether both of the bands of this doublet were JBP1, we blotted our HA-JBP1 immunoprecipitates with antisera generated against JBP1. Both bands of the doublet reacted with this antisera as shown in Fig. 6, suggesting that the lower band of the doublet represents the endogenous JBP1 that co-immunoprecipitated with HA-JBP1. We confirmed the ability of JBP1 to bind to itself with an in vitro binding assay (data not shown). In this assay, the GST-JBP1-N268 protein bound the full-length JBP1 protein produced in COS cells. These data indicate that JBP1 forms homodimeric or multimeric complexes.
Northern Analysis of JBP1-In our initial searches for related proteins in the GenBank data base, we noticed that there were expressed sequence tags corresponding to JBP1 from a wide variety of tissues including those from embryonic and fetal tissue. To determine the tissue expression pattern and transcript size, we probed two human multiple tissue Northern blots with the JBP1 sequence. As can be seen in Fig. 7, the JBP1 mRNA is widely expressed in human tissues as a major transcript of 2.5 kb.
FISH-A combination of FISH and sequential G-banding was used to determine the chromosomal localization of JBP1. In situ hybridization with the biotinylated cDNA probe regionally localized JBP1 to a medium-sized acrocentric chromosome. Lysates from COS-1 cells expressing different Jak kinases were incubated with the GST-JBP1-N268 (indicated as GST-JBP1 in the figure) fusion protein or GST alone (0.8 -3.0 g) immobilized on glutathioneagarose. The proteins associated with GST or GST-JBP1-N268 were resolved with SDS-PAGE, transferred to PVDF membranes, and blotted with anti-Jak1, anti-Jak2, anti-Jak3, or anti-Tyk2 antibodies. I represents 10% of the input COS-1 cell lysate used for the binding assay.
Representative results of hybridization of JBP1 to prepared metaphase spreads are shown in Fig. 8A. Hybridization of JBP1 was specific to the proximal long arm of a group D [13][14][15] (70) chromosome in 10 metaphase spreads analyzed by FISH. A combination of FISH and sequential G-banding of the same metaphase spreads examined for hybridization further localized JBP1 between bands q11.2-q21 on the long arm of chromosome 14 (Fig. 8B). DISCUSSION We used the yeast two-hybrid system to identify new Jak2interacting proteins. One of the proteins identified was a human homologue of the protein encoded by the S. pombe gene skb1 (shk1 kinase-binding protein 1) and that encoded by the S. cerevisiae gene, HSL7 (histone synthetic lethal 7) (47,48). While the defects resulting from the genetic disruption of skb1 and HSL7 implies that these proteins may have a role in regulating the cell cycle or the control of cell morphology, no functional motifs or activities for these proteins were reported (47,48). Comparison of the human protein with homologues in S. cerevisiae (48), S. pombe (47), C. elegans (47), D. melanogaster (this study), and M. musculus (this study) demonstrate that their primary sequence has been well conserved as shown in Fig. 2A. Given the conserved nature of these proteins and their implication in the control of cellular morphology and the cell cycle, we sought to determine a biochemical activity for this newly identified human protein.
Sequence analysis of these proteins revealed homology to protein methyltransferases (Fig. 2B). While the primary sequence of protein methyltransferases are poorly conserved, three motifs have been reported to exist among this diverse family of enzymes (68 -70). The homology between the human protein which we have identified, called JBP1 and protein methyltransferases included these motifs as shown in Fig. 2B. One of these motifs, called motif I, frequently contains the sequence GXGXG (68,69). Motif I is present in many methyltransferases and was shown by crystallography to be within the AdoMet binding pocket of the HhaI DNA methyltransferase (68,69,(77)(78)(79). The conserved GXGRGP region of JBP1 and its homologues likely represents motif I in these proteins and is followed by regions homologous to motif II and III as shown in Fig. 2B.
We first assayed the ability of JBP1 to bind the universal methyl group donor AdoMet with a UVcross-linking assay (Fig.  4A). After demonstrating that JBP1 could bind AdoMet, we tested whether this protein could methylate proteins such as histones, cytochrome c, and myelin basic protein, which are commonly used as methyl group acceptors. HA-JBP1 immunoprecipitates could methylate histone H2A, histone H4, and myelin basic protein. In addition, we also observed the methylation of proteins that co-immunoprecipitated with HA-JBP1 from HeLa cells. We speculate that these proteins likely represent some of the endogenous substrates for JBP1. The ability of JBP1 to bind AdoMet along with the homology between JBP1 and a number of protein methyltransferases suggests that the methyltransferase activity present in our JBP1 immunopre- cipitates is due to JBP1. To test this hypothesis further, we created two point mutants containing substitutions of the central glycine and the arginine of the invariant GXGRGP region conserved among JBP1 and its homologues. Immunoprecipitates from both point mutants exhibited little or no methyltransferase activity, indicating that JBP1 is required for this activity (Fig. 4C). In addition, the protein methyltransferase activity is present when JBP1 is purified by affinity chromatography with an anti-Flag column (data not shown). This purified material readily methylates histones and appears to have an even higher activity for myelin basic protein (data not shown). We are currently unsure if JBP1 is the sole protein responsible for this activity or whether it is a subunit of a protein methyltransferase complex (for reviews, see Refs. 80 and 81). Myelin basic protein and histones have previously been shown to be methylated on arginine residues (64, 65, 74, 82, 83). Based on the homology between JBP1 and protein-arginine methyltransferases and the ability of JBP1 to label proteins known to be arginine-methylated, we hypothesize that JBP1 and its homologues are likely to represent a new group of proteins involved in arginine methylation. Based on its ability to methylate myelin basic protein, JBP1 (and its homologues) may represent the first cloned components of a type II arginine methyltransferase (80,81). Unlike the large number of protein kinases which have been described over the past two decades, it is only recently that cDNAs encoding arginine methyltransferases have been identified and characterized (64,65,75,84).
It is interesting that homologues of JBP1 in S. pombe and S. cerevisiae have been linked genetically and or biochemically to protein kinases. HSL7 disruption in S. cerevisiae results in G 2 arrest, which is Swe1-dependent (48). Skb1 binds to Shk1 and is associated with Cdc2 (49). Our work on JBP1 as well as work on its homologues in yeast suggest that there may be a link between protein kinases and this conserved family of proteins. While the link between protein kinases and protein methylation is unclear, protein methylation may add another mechanism by which complex cellular events are controlled. It is possible that involvement of JBP1 may be one mechanism through which the Jaks exert their influence on the cell cycle.
Methyltransferases represent a diverse family of enzymes which can transfer methyl groups from AdoMet to a variety of substrates including nucleic acids, small molecules and proteins (85,86). Protein methylation reactions can vary with respect to the site of methylation and the nature of the covalent bond formed. Methylation of the ␣-amino group of certain NH 2terminal amino acids as well as the N-methylations of histidine, lysine, and arginine residues are irreversible, and are likely to play a structural role (70,85,86). Other methylations occurring at carboxyl groups are reversible and may therefore be involved in more dynamic processes (86 -89). Protein meth-  6. Homodimerization of JBP1. HA-JBP1 was immunoprecipitated from HeLa cells stably transfected with pEF2-HA-JBP1. Immunoprecipitates were washed and eluted proteins separated by SDS-PAGE. After proteins were transferred to PVDF membranes, the membranes were Western blotted with antibodies against HA or JBP1. After transfer, the remaining proteins were visualized by staining with Coomassie Blue.
FIG. 7. Northern analysis of JBP1 in human tissues. Blots containing oligo(dT)-selected mRNA from multiple human tissues (CLON-TECH) were hybridized with labeled DNA corresponding to the sequence encoding JBP1. After hybridization and washing, autoradiography was performed which revealed the presence of a major transcript between 2.4 and 2.6 kb. Molecular weight standards are indicated on the left of each panel. The tissues are as follows: H, heart; Br, brain; Pl, placenta; Lu, lung; Li, liver; SM, skeletal muscle; K, kidney; Pa, pancreas; Sp, spleen; Th, thymus; Pr, prostate; Te, testicles; Ov, ovaries; SI, small intestine; LI, large intestine; Le, peripheral blood leukocytes.
yltransferases have been implicated in the repair of damaged proteins, and knockout of one such protein-repair enzyme in mice results in the accumulation of altered proteins, retardation of growth, and fatal seizures (80,86). Despite the fact that protein methylation has been known for some time, and a role for protein methylation in cellular signaling has been the subject of a recent review, the exact role of this modification in many processes is still not well understood (85,86,90). However, some interesting observations have been recently reported. Arginine methylation was shown to facilitate the export of certain hnRNPs (heterogeneous nuclear riboproteins) from the nucleus (91). Another report demonstrated binding of a protein arginine methyltransferase to the IFNAR1 chain of the IFN-␣ /␤ receptor complex and experiments with antisense oligonucleotides provided evidence that this methyltransferase is involved in mediating the antiproliferative affect of IFN-␣/␤ (75). Recently, a 72-kDa protein, identical to JBP1, was shown to bind to pICln (I ϭ current, Cl ϭ chloride, n ϭ nucleotidesensitive), a protein lacking a definitive function that was proposed to be a cytosolic regulator of a swelling-induced chloride channel (92). No activity or functional homology was reported for this 72-kDa protein (IBP72 for 72-kDa pICln-binding protein); however, its interaction with a protein speculated to be involved in signaling is intriguing (92). The sequence homol-ogy between JBP1 and its homologues suggests that these proteins share similar function. Indeed, the human homologue of Skb1 was shown to functionally complement skb1 in S. pombe and Hsl7p in S. cerevisiae (49). 3 Now that a biochemical activity has been identified for JBP1, additional experiments can be designed to help elucidate the role of this protein and its homologues in cellular signaling, the cell cycle and the control of cell morphology.
Note Added in Proof-Fujita et al. (Fujita, A., Tonouchi, A., Hiroko, T., Inose, F., Nagashima, T., Satoh, R., and Tanaka, S. (1999) Proc. Natl. Acad. Sci. U. S. A. 96, 8522-8527) recently reported that the homologue of JBP1 in S. cerevisiae, Hs17p, interacts with the p21 Cdc42/Rac -activated kinase (PAK) Ste20 and appears to function as a negative regulator of Ste20p in the filamentous growth-signaling pathway. Although they reported no functional homologies or enzymatic activities for Hs17p, their results in combination with our discovery that JBP1 exhibits methyltransferase activity suggests that methylation may be involved in this regulation.